| *** charlesw has joined #magnetodb | 00:28 | |
| *** charlesw has quit IRC | 02:33 | |
| *** rushiagr_away is now known as rushiagr | 03:56 | |
| *** rushiagr is now known as rushiagr_away | 04:43 | |
| *** rushiagr_away is now known as rushiagr | 05:16 | |
| *** ajayaa has joined #magnetodb | 05:47 | |
| openstackgerrit | Dmitriy Ukhlov proposed a change to stackforge/magnetodb: Improve json schema validation https://review.openstack.org/121162 | 09:57 |
|---|---|---|
| *** rushiagr is now known as rushiagr_away | 10:09 | |
| ajayaa | Hi isviridov, dukhlov | 11:14 |
| ajayaa | Is anyone looking into celiometer integration with mdb? | 11:15 |
| ajayaa | I see that there is a blueprint for celiometer and there is no assignee. | 11:16 |
| ajayaa | https://blueprints.launchpad.net/magnetodb/+spec/celiometer-integration | 11:16 |
| *** rushiagr_away is now known as rushiagr | 11:23 | |
| dukhlov | hi, ajayaa | 11:48 |
| dukhlov | isviridov knows I guess | 11:49 |
| ajayaa | dukhlov, hi | 11:49 |
| dukhlov | I know that we have started this activity | 11:49 |
| dukhlov | but I don't know it's current status | 11:49 |
| ajayaa | dukhlov, I think someone is working on the notification part. I see some code related to it in the repo. | 11:51 |
| isviridov | ajayaa, hello | 12:14 |
| isviridov | About celiometer integration | 12:14 |
| ajayaa | hi isviridov, | 12:15 |
| isviridov | Yes we have started it, I mean added notifications for mostly all events. | 12:15 |
| isviridov | I mean #link https://github.com/stackforge/magnetodb/blob/master/magnetodb/notifier/__init__.py | 12:16 |
| isviridov | Another part of this is monitoring-api https://blueprints.launchpad.net/magnetodb/+spec/monitoring-api what will be used for collecting data about mdb usage by celiometer. | 12:18 |
| isviridov | That is | 12:19 |
| isviridov | ajayaa, do you have usecases with celio? | 12:28 |
| ajayaa | isviridov, I was going through amazon's apis and realized we don't have anything for measuring usage info in mdb. | 12:44 |
| ajayaa | sorry, I was out for some time. | 12:44 |
| ajayaa | isviridov, and thought to ask you whether anyone is working in this area. | 12:45 |
| isviridov | ajayaa, good aproach! | 12:51 |
| isviridov | So, the closest person to it now ominakov. Notifications was implemented by CharlesWang | 12:51 |
| isviridov | About plans and roadmap you can always contact me. | 12:52 |
| isviridov | But some ideas about usage statisticsL | 12:52 |
| ajayaa | isviridov, sure. | 12:52 |
| isviridov | In DynamoDb we have it per request, and it is clear that we can't calculate it each data request. | 12:53 |
| ajayaa | Why not? Each time a user does a read we can send a notification to celiometer who will do the book-keeping. | 12:54 |
| isviridov | It sounds like dependency on celiometer | 12:55 |
| isviridov | Do you suggest to get usage data from celiometers and return it via some of mdb API? | 12:55 |
| ajayaa | Yes. | 12:56 |
| isviridov | It IS dependency on celiometer. | 12:56 |
| ajayaa | There will be a dependency with celiometer I agree. But many of the OS components are interdependent. | 12:56 |
| isviridov | Could you imagine that according to SLA mdb must response in 20ms. | 12:57 |
| ajayaa | Or we can do the book-keeping ourselves in mdb with the same backend. | 12:57 |
| isviridov | But there is interdependent component celiometer and it has another SLA. And could respond in 2sec | 12:58 |
| isviridov | But let me continue | 12:58 |
| ajayaa | isviridov, I agree. We can't gurantee response time of celiometer. | 12:58 |
| isviridov | In mdb we have introduced monitoring API to separate access rights between monitoring system (celiometer, collectd, whatever else) and data api (clients, horizon, magnetodb-cli) | 12:59 |
| isviridov | Another effect of it, we can provide different SLA for such services. | 13:00 |
| isviridov | The only question is DynamoAPI. It has usage statistics in data api built-in. I think we can implement some kind of internal data structure and execution routine to update it. | 13:01 |
| isviridov | In this case we will be able to response via DynamoDB API fast and robust. | 13:01 |
| openstackgerrit | Dmitriy Ukhlov proposed a change to stackforge/magnetodb: Improve json schema validation https://review.openstack.org/121162 | 13:01 |
| isviridov | That is. What do you think ajayaa? | 13:01 |
| ajayaa | isviridov, sounds like a good plan. :) | 13:02 |
| ajayaa | isviridov, But if we anyway are going to maintain internal DS for dynamodb apis, then why do we need to do separate data api and usage api in OS apis? | 13:03 |
| isviridov | ominakov-, is working on collecting data from cassandra via JMX (thank you for hint and link to article) we will see how fast and robust it is. | 13:03 |
| isviridov | ajayaa, because of different users. Monitoring components will not have access to data itself. | 13:04 |
| ajayaa | The point I am trying to make is: We are going to store usage info in internal structures for dynamodb apis. Why no use the same internal structures for openstack apis in mdb and expose them as mdb apis? | 13:07 |
| isviridov | ajayaa, got you. It is really possible and probably can be next iteration of implementation. | 13:09 |
| isviridov | ajayaa, another reason, we are not concentrated on DynamoDB and its performance right now. So, just moving out the monitoring works good for us. | 13:11 |
| ajayaa | isviridov, okay. One thing that is not clear to me is: How would the user get to know his usage info? through monitoring apis or celiometer apis? | 13:13 |
| isviridov | ajayaa, monitoring api and all systems above it. It can be celiometer or anything else like graphite, nagios depends on deployment. | 13:17 |
| isviridov | ajayaa, and sure CLI | 13:17 |
| ajayaa | isviridov, Thanks. :) | 13:19 |
| * isviridov has meeting time | 13:20 | |
| *** isviridov is now known as isviridov_at_met | 13:20 | |
| *** charlesw has joined #magnetodb | 13:27 | |
| openstackgerrit | Aleksey Chuprin proposed a change to stackforge/magnetodb: (WIP) Investigation problem with tox job timeout https://review.openstack.org/119467 | 14:50 |
| *** ajayaa has quit IRC | 14:53 | |
| openstackgerrit | Dmitriy Ukhlov proposed a change to stackforge/magnetodb: Improve json schema validation https://review.openstack.org/121162 | 15:31 |
| openstackgerrit | Andrei V. Ostapenko proposed a change to stackforge/magnetodb: (WIP) "ALL_OLD" return_values for put_item https://review.openstack.org/121911 | 15:41 |
| *** rushiagr is now known as rushiagr_away | 15:43 | |
| *** isviridov_at_met is now known as isviridov | 20:01 | |
| openstackgerrit | Aleksey Chuprin proposed a change to stackforge/magnetodb: (WIP) Investigation problem with tox job timeout https://review.openstack.org/119467 | 20:19 |
| *** isviridov is now known as isviridov_away | 20:35 | |
| *** charlesw has quit IRC | 22:32 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!